NVIDIA has begun deploying its Spectrum-6 Ethernet switch, which boasts a capacity of 102.4 terabits per second, across leading AI factories worldwide, including those operated by CoreWeave, Microsoft, Nebius, SpaceXAI, and Tesla. This switch is designed to enhance performance for large-scale AI clusters by minimizing congestion and latency during training workloads. The launch aligns with a growing trend in the industry towards ‘AI factories,’ which are integrated data centers focused on the ongoing training and deployment of AI models, reflecting a broader transition to Ethernet-based interconnects in next-generation AI supercomputers.

Tesla: Tesla is an automotive and energy company heavily invested in AI for autonomous driving, robotics, and large‑scale training clusters. Tesla is listed as an early user of NVIDIA Spectrum‑6, incorporating the switch into its AI factory networks to support increasingly demanding training and inference tasks.
NVIDIA: NVIDIA is a leading semiconductor and computing platform company that designs GPUs and high‑performance networking systems widely used for AI and data center workloads. In this news, NVIDIA is introducing the Spectrum‑6 Ethernet switch, a high‑capacity component of its AI networking stack that is being deployed in large‑scale AI factories and the Vera Rubin platform.
Nebius: Nebius is a cloud platform designed to provide high‑performance compute and storage services with a focus on AI and data‑intensive workloads. Here, Nebius appears as one of the initial cloud operators integrating NVIDIA Spectrum‑6 switches to strengthen the network fabric of its AI production environments.
SpaceXAI: SpaceXAI refers to AI infrastructure efforts associated with SpaceX, leveraging high‑performance compute for autonomous systems, simulation, and related workloads. In this news, SpaceXAI is named among the first organizations deploying NVIDIA Spectrum‑6 to accelerate the network layer of its AI factories.
CoreWeave: CoreWeave is a specialized cloud provider focused on GPU‑accelerated infrastructure for AI, VFX, and other compute‑intensive applications. In this context, CoreWeave is one of the first operators adopting NVIDIA’s Spectrum‑6 switches to scale and accelerate its AI factory networking backbone.
Microsoft: Microsoft is a global technology company that develops software, cloud services, and AI platforms, including Azure for large‑scale AI training and inference. In the news item, Microsoft is highlighted as an early adopter of NVIDIA Spectrum‑6 to enhance the networking performance of its gigascale AI infrastructure.

AI_Networking: Recent industry coverage emphasizes that NVIDIA’s Spectrum family is designed specifically to optimize Ethernet fabrics for large‑scale AI clusters, reducing congestion and latency in training workloads.
Ethernet_for_AI: Technical deep dives published in the last few weeks note that Spectrum‑6 class switches offer dense high‑speed port configurations, aligning with a broader shift toward Ethernet‑based interconnects in next‑generation AI supercomputers.
AI_Factories_Trend: Over the past month, multiple cloud and AI infrastructure providers have highlighted the concept of ‘AI factories’ to describe vertically integrated data centers optimized for continuous AI model training and deployment.